The transistor, a superior new means of controlling the flow of electrons, has now been developed in a practical form; it liberates electronics from the limitations of the vacuum tube.
By Louis N. Ridenour
Electronics was born 45 years ago when Lee De Forest invented the first three-electrode vacuum tube—the "audion." The two world wars gave electronics the impetus that made it the huge industry it is today. World War I, during which radio telephony was developed from a laboratory curiosity into a reasonably practical means of communication, gave birth to commercial radio: the first broadcasting station, KDKA, was licensed in 1920. World War II, which opened a new horizon in electronics through the development of radar and its attendant technology, produced television. There had been experiments with television before the war, but it was the wartime advances achieved on behalf of radar that made commercial TV immediately realizable. The story of television's postwar rise is dramatically told in production figures: the number of television receiving sets made in the U. S. rocketed from 6,500 in 1946 to 7,500,000 in 1950. Besides television, World War II produced another major electronic advance: high-speed digital computing machines, which also seem destined to have a profound social impact, though this is still largely in the future.
Paradoxically the vacuum tube, which made all this possible, has now become a bottleneck impeding further progress in electronics. The electronics art has reached a stage where a fundamental new development is needed to rescue it from the limitations of the vacuum tube. Just such a development has recently arrived on the scene. It is the transistor, which can perform most of the functions of the vacuum tube and escapes most of its limitations. The transistor promises to revolutionize electronics; indeed, the revolution is already beginning.
To understand the causes and nature of this revolution, we must first look into what is wrong with the vacuum tube. To begin with, the vacuum tube is inherently rather short-lived and unreliable. This difficulty has been exacerbated by the endemic price competition in the commercial electronics industry. Competing on a price basis for the mass market for radio and television sets, manufacturers have achieved prodigies of economy in design and of efficiency in production. But this has exacted its inevitable toll in quality. Most commercially available vacuum tubes are not as reliable as they could be if more care, meaning more money, were put into their design and manufacture. This shortcoming is inconvenient enough in a 6-tube or 8-tube radio set; it is considerably more inconvenient in a television receiver using 25 to 35 tubes, which quadruples the chances of set failure for a given rate of tube failure. As electronic equipment becomes more complex, the reliability of individual components becomes more and more important.
The telephone system affords an excellent illustration. A transcontinental phone call requires the proper operation of 12,300 vacuum tubes, 112,000 resistors, 97,000 capacitors and large numbers of other components. Hence in the telephone system the reliability of electronic equipment is far more important than its initial cost; a cheap device that requires much maintenance will be much more expensive in the end than a more carefully built apparatus that can work for years with no attention. Because of its extreme concern with reliability, the telephone system has gone to great pains to make the components of its electronic equipment as dependable and long-lived as the state of the art will permit. For the Bell Telephone System the Western Electric Company manufactures equipment to exacting specifications worked out by the Bell Telephone Laboratories. The vacuum tubes it produces are the most durable in the world. In a submarine cable recently laid on the bottom of the sea between Key West and Havana are six repeater amplifiers, each containing three vacuum tubes. They are of course inaccessible for replacement. But if any of those 18 tubes sealed into the cable should fail in the next 20 years, the designers of the cable will be surprised.
Yet there are limits to the reliability that can be built into vacuum tubes even with great care, ingenuity and cost. The telephone system has never been able to use electronic equipment for two important functions—central-office switching and signal amplification on local lines—partly because no vacuum tube is sufficiently dependable.
Reliability in electronic equipment is especially important to the military services. The cost of field maintenance of their electronic gear at present is about 10 times the initial cost of the equipment. The Army, Navy and Air Force are using increasingly complicated electronic devices (e.g., radar bombsights) in increasing numbers and putting increasing reliance on them; the Navy, for example, has a bomber aircraft that is not built for an optical bombsight at all—even has no transparent window for one. In the Air Force's B-36 the electronic equipment accounts for about 10 per cent of the total cost of that very expensive airplane, and in the newest all-weather fighters electronic devices make up 20 per cent of the cost of the plane. Hence to the military, as well as to civilians, the reliability of electronic equipment is a matter of urgent concern.
A further problem in electronic equipment is that of size. As the equipment grows in complexity, it grows more and more bulky. This has led to an attempted solution rather inelegantly called "miniaturization." Vacuum tubes have been designed in miniature and sub-miniature models with essentially the same electrical capabilities as the old standard sizes. Resistors, capacitors, transformers and other circuit elements have been greatly reduced in total volume by radical redesign. "Printed circuits" and other compact schemes for prefabricated wiring have been devised. And the utmost attention has been paid to "designing the air out of" the equipment as a whole: every nook and cranny in the chassis is filled with wiring and components.
One example of this design trend is…a radio range receiver before and after miniaturization. The volume of the set of standard design is about 300 cubic inches; that of the miniature set is 55 cubic inches—a sixfold reduction. And this was done in the case of a set which had been rather compactly designed, in terms of conventional materials and construction, to begin with.
There is no doubt of the usefulness of such miniaturization for equipment that must be stowed in a small space. But miniaturization intensifies one of the outstanding deficiencies of the vacuum tube: namely, its heavy power requirement and the large amount of heat dissipated in the tube. The vacuum tube, a grandchild of the incandescent lamp, uses heat produced by electric power to boil electrons off the cathode. Generally speaking miniaturization does not reduce the power requirement: the miniature version of the range receiver…takes just about as much power as the larger model—some 30 watts. In the larger model, with a volume of 300 cubic inches, power is liberated at the rate of a tenth of a watt per cubic inch; the components are reasonably well separated from one another, so that ventilation is easy, and the temperature rise of the equipment in operation is modest. On the other hand, in the small model, with components packed as tightly as possible, power is dissipated at the rate of more than half a watt per cubic inch, which is the average rate of power expenditure in the ordinary home electric oven. As a result some spots inside the set reach temperatures as high as 400 degrees Fahrenheit.
This high operating temperature, which is typical of most miniaturized electronic equipment, raises a host of new problems. Connections must be made with special high-melting-point solder. Temperatures must be kept below the softening point of the glass in the tube envelope. And the high temperature greatly reduces the operating life of most of the tube's components.
In short, with the increasing complexity of electronic equipment the relatively low reliability of the vacuum tube and its prodigal use of electric power have become major difficulties.
Thus nearly half a century after De Forest's invention of the tube that gave birth to electronics we are brought to the conclusion that there is nothing wrong with electronics that the elimination of vacuum tubes would not fix!
Until 1948 there seemed no way out of this difficulty. But in that year John Bardeen, W. H. Brattain and William Shockley of the Bell Telephone Laboratories announced the invention of the transistor. At the time the device looked promising but had important practical shortcomings. Transistors could not be produced reliably and with predictable electrical characteristics. They were "noisy," would not operate at frequencies higher than a few megacycles per second and had other deficiencies. But three years of hard work at the Bell Laboratories have removed most of the early doubts and objections. Transistors can now be made so that their performance is within 20 per cent of specifications, which is as close production control as has been achieved in half a century of experience in making vacuum tubes. It is likely that further development will lead to methods for producing transistors that conform to specifications within 10 or even 5 per cent.
The transistor is a device made of a crystalline semiconducting material, usually germanium. In this material the number of atomic electrons in the crystal does not quite match the number needed to produce a normal crystal lattice. If there are a few excess electrons, these can travel through the crystal as readily as free electrons traverse the empty space inside a vacuum tube. If there is a deficiency of electrons, the "holes" representing vacant electron sites also can travel freely through the crystal. At barriers between electron-rich and electron-deficient regions of the germanium, the flow of electrons or holes can be controlled by an applied signal in much the way that the plate current of a vacuum tube is controlled by the voltage applied to the grid. Like the vacuum tube, the transistor can control a larger power than is applied to it—that is, it can amplify.
The important practical difference between the transistor and the vacuum tube is that no power whatever is needed to set loose the free electrons or holes in the transistor. There is no need to "boil" electrons out of the solid material; instead, the electrons are controlled as they move inside the solid.
It appears that the durability of transistors, and therefore the reliability of circuits using them, will be excellent. The earliest transistors made had average lifetimes in the vicinity of 70,000 hours, already several times greater than the lifetime of a vacuum tube, and present units are much longer-lived. In principle a properly made transistor should last forever if it is not abused.
Moreover, transistors are highly resistant to mechanical shock and vibration. They survive very severe shock tests without damage. Everyone knows that tapping on the tubes of a radio set produces an unpleasant ringing. This is due to changes in the electrical properties of the vacuum tubes, produced by the mechanical vibration of their elements. Transistors can withstand high accelerations without being subject to such "microphonic" effects. And in the newer transistors the internal electrical "noise" has been reduced to a level that compares quite favorably with that of vacuum tubes.
A large part of the improvement in the performance of the device is due to the development of a new design called the "junction transistor." The early units consisted of a germanium crystal touched by two closely spaced fine wires—"cat's whiskers." In the junction transistor this point-contact arrangement has been replaced by a large area contact. A thin layer of electrically positive (electron-deficient) germanium is sandwiched between two electrically negative (electron-rich) ends. The transistor action now takes place over the whole area of contact (junction) between the two types of germanium. It therefore operates more efficiently and consumes far less power.
Obviously the transistor's most important advantage is the modesty of its power requirements. In the first place, it is ready for instant operation without the lengthy warm-up needed by a vacuum tube, so that power does not have to be applied constantly, as it must in vacuum-tube circuits, to equipment that must be ready to operate instantly on demand. Secondly, the total operating power used in a transistor device may be incredibly small—about one-millionth of the power required by a similar vacuum-tube circuit. Ralph Bown, director of research at the Bell Telephone Laboratories, has dramatized this property of the transistor by a vivid analogy. He points out that to amplify the average signal in most electronic equipment, which is in the neighborhood of a microwatt (one-millionth of a watt), a vacuum-tube system requires the expenditure of a watt of power; this is like sending a 12-car freight train, complete with locomotive, to get a pound of butter. The tiniest hearing-aid tubes, in which electrical performance has been considerably degraded in favor of low powerdrain, use about 20 milliwatts. In terms of Bown's simile, this is like sending a 10-ton truck for one pound of butter. The transistor, on the other hand, needs only a microwatt to handle a microwatt.
This million-fold decrease in power requirement is so great that it is hard to appreciate. In a hearing aid a transistor can produce a sound amplification of 45 or 50 decibels at one stage, in contrast to the 25 decibels that can be achieved with low-power-drain vacuum tubes. Yet a transistor hearing aid could run on one set of batteries for about three years, whereas a vacuum-tube hearing aid uses up a set of batteries in a few days. A single transistor amplifier stage could be powered with the energy a flea would expend if it jumped once every eight seconds. We are dealing here not with flea power but rather with lazy-flea power!
In electronic equipment using transistors the problem of high temperature, so serious in the case of vacuum tubes, vanishes altogether. The power required is so small that heating is infinitesimal. The miniaturization of electronic equipment now makes solid sense. The transistor is far more compact than any vacuum tube. A typical transistor occupies only about a two-thousandth of a cubic inch, whereas the smallest subminiature vacuum tube is about one eighth of a cubic inch in volume.
Even at the present very early stage of transistor development it seems certain that transistors will replace vacuum tubes in almost every application. What results can we expect from this major revolution in the techniques and capabilities of electronics? Since the revolution is just beginning, we can only speculate.
We can begin by noting the great gain that electronic machines represent over machines of the strictly mechanical type—the distinction, let us say, between an electronic device such as a television receiver and a mechanical contrivance such as a cash register. The most meaningful difference between electronic machines and mechanical machines is in the level of complexity they can attain. A century ago the ingenious English mathematician Charles Babbage designed an analytical engine which in principle was much the same as the large-scale digital computing machines of today. But Babbage found it impossible to build his proposed machine; it was too complicated to be within the capabilities of an all-mechanical design. The realization of devices of this complexity, including high-speed digital computers and long-distance telephone circuits, became possible only when electronics was developed.
Generally speaking, vacuum-tube electronics permits the building of machines about a hundred times more complicated—and therefore at least a hundred times more competent—than the most complex mechanical machine it is practical to make. Electronic machines enjoy another advantage over mechanism: they can work much faster. The typical unit operation in an electronic machine is accomplished in a time of the order of a millionth of a second; the typical unit operation of a mechanical machine takes about a thousand times longer.
But when we get to such a complex electronic machine as the modern digital computer, the shortcomings of the vacuum tube itself become troublesome. The power the vacuum tube requires and its relatively short lifetime pose substantial obstacles to any further increase in complexity. Now the transistor promises to remove these obstacles. Its lifetime is surely long and may be indefinite. Its power requirement is a million times smaller than that of the vacuum tube. Its electrical performance is at least as good, and in many ways better.
All these favorable properties of the transistor suggest that its use as a substitute for the vacuum tube in electronic machines may permit such machines to grow to levels of complexity now unattainable. Just as vacuum-tube electronics permitted a hundred-fold increase in complexity over that permitted by mechanism, at a conservative estimate transistor electronics may allow a hundred-fold increase in complexity over vacuum-tube machines.
From the general philosophical standpoint, it is of interest to compare the most complicated machine we can make with the machines we observe in nature, such as the human central nervous system. Warren McCulloch of the University of Illinois Medical School has done this in very entertaining terms. He finds that the Eniac computer, containing about 10,000 basic on-or-off elements, is a million times less complex than the brain, which has 10,000 million neurons. The Eniac, indeed, has about the complexity of the nervous system of the flatworm. It has one advantage: its unit operations are accomplished about a thousand times faster than are the unit operations of the brain. Thus if we made a sort of figure of merit for comparing the competence of man-made and natural machines, taking into account both complexity and speed, we should find the Eniac—for those operations fitted to its very low complexity—only about a thousand times less competent than the human brain.
McCulloch remarks that if we made a vacuum-tube computer as complex as the brain, it would require a skyscraper to house it, the power of Niagara to operate it and the full flow of water over the falls to keep it cool. This is altogether a criticism of vacuum tubes. If, as seems reasonable to suppose, the use of transistors will permit a further hundred-fold increase in the complexity of our machines, we shall be able to build, in no greater space and with smaller power requirements than are needed now for vacuum-tube computers, a device only 10,000 times less complicated than the brain. Since it will work a thousand times faster, such a transistor device may be, for those jobs to which its low complexity suits it, as much as one-tenth as competent as the human brain.
This is an exciting prospect, but it has not yet been achieved. Curiously, its achievement seems to rest on the elimination from electronics of the vacuum tubes which gave electronics birth.